Column
Agent Engineering
Five essays from context foundations to production systems
This column traces one complete path: from principles, to practice, to reflection.
The first essay lays the foundation — context is not prompt, and why context engineering is becoming the new bedrock of AI applications. Every architectural decision that follows builds on this layer.
The second opens the panorama — where the 98.4% of agent engineering actually lives. The model is just 1.6%; the rest is harness.
The third and fourth are two sharply contrasting case studies: open-lovable tames bare APIs without any agent framework, while Relay is a production job-hunting agent system designed from scratch. One subtracts; the other adds.
The final essay steps outside engineering — when agents gain memory and continuity, where does the “self” begin? From Locke’s personal identity to OpenClaw, philosophy and engineering look each other in the eye.
Reading 1 → 5 in order works best, but every essay stands on its own.
Contents
Best read in orderClaude Tag Deep Dive: From Shared Slack AI to an Organizational Agent Runtime
Claude Tag turns Slack into a shared agent runtime with identity, memory, and tools. This deep dive covers its architecture, risks, Chinese peers, and future.
Open Design 0.16.1: A Design Harness for Coding Agents
A version-pinned guide to Open Design 0.16.1: its four-plane architecture, local and cloud paths, BYOK privacy boundaries, SSRF controls, and practical fit.
Designing devbox-doctor: A Safer Mac Toolchain Audit
A design for auditing Mac developer tools with read-only evidence, bounded AI judgments, privacy safeguards, graceful fallbacks, and false-positive tests.
Agent Skill Design: What a Dangerous SKILL.md Taught Me
Agent Skill and SKILL.md design through a storage-cleanup case study: architecture, permission boundaries, prompt contracts, real-machine tests, and lessons.
Inside open-lovable: Search, Text Protocols, and Sandboxes
A commit-pinned audit of open-lovable’s text protocol, Agentic Search, Morph edit path, and E2B/Vercel sandbox boundaries for practical AI app builders.
Relay Agent Architecture in 2026: A Local Implementation Audit
A source-based audit of Relay’s coordinator, five domain agents, LangGraph HITL, browser delivery, durability gaps, and remaining 2026 migration debt.
Context Engineering: The New Foundation for AI Agents
Learn how context engineering improves AI agents through retrieval, compaction, memory, and token budgeting—and why high-signal context beats a larger window.
Agent Engineering Harness: The Eight Pillars Behind the 98.4%
Agent Engineering Harness explained through eight production pillars: orchestration, context, memory, tools, reliability, evaluation, cost, and governance.
Agent Identity: From Locke to OpenClaw
A practical framework for continuous AI agent identity, covering memory, permissions, provenance, multi-agent handoffs, and cross-session evaluation design.